Mutations in RNU7-1 Weaken Secondary RNA Structure, Induce MCP-1 and CXCL10 in CSF, and Result in Aicardi-Goutières Syndrome with Severe End-Organ Involvement

We investigated 3 patients with AGS from unrelated non-consanguineous families of European origin (full clinical descriptions are given in supplemental data S1). All 3 patients presented with classical features of AGS including psychomotor retardation, spastic quadriparesis, and intracranial calcifications (Fig. 1a). In addition to calcifications, ultrasound imaging revealed lenticulostriatal vasculopathy in P1 (Fig. 1a). P1 and P2 are in clinical follow-up and currently 2 and 3 years old, respectively. At 19 years of age, P3 developed tonic–clonic convulsions. Laboratory analysis of P3 revealed mild hemolytic anemia and severe renal insufficiency (creatinine of 2.85 mg/dl, eGFR 30 ml/min) with arterial hypertension. Following palliative care, he died 1 month later. Histological examination of the brain demonstrated calcifications in the walls of arterioles and capillaries, demyelination of neurons, and gliosis (Fig. 1b). Perivascular clustering of histiocytes and CD68 + microglial cells could be observed (Fig. 1b). Histopathology of the kidney showed fibrin thrombi and intimal proliferation, hallmarks of thrombotic microangiopathy (Fig. 1b).

WES in these 3 patients did not reveal pathogenic variants in the seven well-known AGS-related genes nor in the recently described LSM11 gene (see Table S2 for 20 × coverage). Subsequently, we performed Sanger sequencing and found biallelic mutations in RNU7-1 (NR_023317.1), encoding U7 snRNA, in these 3 patients. All patients (P1, P2, and P3) carried an 8 bp deletion in RNU7-1 (n.40_47del), which has been recently associated with AGS [7]. In addition, we identified a novel mutation in RNU7-1 absent in the Genome Aggregation Database (gnomAD); P1 and P3 carried a n.27dup mutation in trans. In P2, we detected a n.40C > G mutation in trans, which has been reported only once in Gen egnomAD. In conclusion, Sanger sequencing revealed that P1, P2, and P3 (postmortem analysis) were carrying compound heterozygous mutations in RNU7-1 (Fig. 1c, d).

The n.27dup mutation (P1, P3) is located in the Sm binding site, the n.40C > G mutation (P2) and the n.40_47del mutation (P1, P2, P3) in the 3′ stem-loop of U7 snRNA (Fig. 2a, b) [7]. The Sm binding site is crucial for the correct assembly of the U7 snRNP [9]. Furthermore, mutations at positions n.23, n.28, and n.30, all located within the Sm binding site, were identified as pathogenic [7]. As such, the patient-derived n.27dup mutation was predicted to impair U7 snRNP assembly. The assessment of pathogenicity of variants in non-protein encoding genes, such as RNU7-1, is challenging. It is not possible to infer damaging status (such as by recognizing a variant as null) and commonly used in silico algorithms are unavailable. To evaluate the structural impact of the 3′ stem-loop mutations of our patients, we performed secondary structure analysis using the RNAfold web server and calculated the minimum free energy (MFE) in which a more negative free energy value represents a more stable structure [10]. All predicted secondary structures containing the previously reported and the newly identified 3′ stem-loop mutations had a more positive MFE and thus reduced stability compared to the consensus sequence (Fig. 2c, Fig. S3). In P2, the RNU7-1 allele containing the n.40C > G mutation also harbored a n.52C > T polymorphism, which was predicted to further destabilize the 3′ stem-loop structure in an additive manner (Fig. 2b, c). The average MFE of homozygous 3′ stem-loop variants in RNU7-1 reported in gnomAD was significantly more negative compared to the average MFE in patients harboring RNU7-1 mutations, suggesting a more stable secondary structure (Fig. 2c, d). In line with the autosomal recessive nature of RNU7-1-mediated AGS, the heterozygous RNU7-1 variants reported in gnomAD spanned the whole spectrum, containing both neutral variants and destabilizing mutations (Fig. 2d, e). Severely destabilizing heterozygous variants (defined as < 2SD of average MFE of homozygous variants or >  − 10.81) were rare (Fig. 2e) with a number of mutant alleles (MFE >  − 10.81) in the population database of 205 out of 152,312 sequenced alleles (gnomAD v3.1) or an estimated 0.135% mean allele frequency (MAF). Some AGS patients harbored RNU7-1 stem-loop mutations with a MFE <  − 10.81 (Fig. 2e). Notably, these less destabilizing mutations were always compound heterozygous with mutations in the Sm binding site or stem-loop mutations with a MFE >  − 10.81 (Fig. 2e).

Next, we set out to assess the functional consequences of the identified RNU7-1 mutations. The U7-snRNP complex is involved in RDH pre-mRNA processing. Impaired assembly of U7 snRNP disrupts endonucleolytic cleavage of the 3′ end of RDH pre-mRNAs, preserving the poly(A) tail. We performed RT-qPCR using oligo(dT) primers which anneal to the poly(A) sequences of aberrant RDH mRNA isoforms. In line with previous findings [7], we observed increased misprocessed polyadenylated mRNAs of both linker (HIST1H1C) and core (HIST1H2AC) RDH in patient fibroblasts (P1, P2, P3) as well as in whole blood (P1, P2) (Fig. 2f, g). Normal protein expression of LSM11, an essential component of the U7-snRNP, was confirmed in primary fibroblasts (Fig. S4). Expression of replication-independent histone mRNAs, which are normally polyadenylated and do not require the U7 snRNP complex, was unaffected (Fig. 2h, i).

We examined an IFN-I signature by quantifying the mRNA expression of 24 interferon-stimulated genes (ISGs) measured on a NanoString platform in whole blood of our AGS patients compared to HCs (Fig. 3a, see “Methods”). Both P1 and P2 demonstrated a mildly elevated ISG score (respectively 3.86 and 3.94) in contrast with a control group of three AGS patients with biallelic mutations in SAMHD1 (7.94), RNASEH2B (10.14), or TREX1 (11.60) (Fig. 3b, Table S5, case descriptions in supplemental data S1). In line with the pronounced IFN-I signature in SAMHD1, RNASEH2B, and TREX1 patients, increased protein concentrations of IFN-a2 could be readily identified in the serum of these patients, but not in our RNU7-1-mutated AGS patients (Fig. 3c).

We quantified IFN-I-induced cytokines in serum of AGS patients. Correlating with the high ISG score and serum IFN-α2 levels, we observed pronounced CXCL10 levels in the serum of patients harboring RNASEH2B (2663 pg/mL) and TREX1 (4010 pg/mL) mutations, whereas the concentration in both RNU7-1-mutated patients (P1: 367 pg/mL, P2: 194 pg/mL) was comparable to HCs (median 232 pg/mL, IQR 66–546) (Fig. 3d). In parallel, we performed cytokine profiling on cerebrospinal fluid (CSF) of RNU7-1 patients (P1, P2). IL-1b, IL-6, and IL-18 could not be detected (data not shown). Low concentrations of IL-1RA (P1: 10 pg/mL, P2: 19 pg/mL), produced by monocytes upon IFN-I stimulation [11], were present in the CSF of patients with RNU7-1 mutations but undetectable in HCs (Fig. 3e). Evaluation of the IFN-I-inducible cytokines monocyte chemoattractant protein 1 (MCP-1 or CCL2) (P1: 644 pg/mL, P2: 1263 pg/mL) and CXCL10 (P1: 2274 pg/mL, P2: 6089 pg/mL) revealed a remarkable increase in the CSF compartment (Fig. 3e). The IFN-g-responsive cytokine CXCL9 could not be detected (Fig. 3e).

Finally, we interrogated STAT1/2 phosphorylation in CD14⁺ monocytes upon ex vivo stimulation of peripheral blood mononuclear cells (PBMCs) with IFN-α2, IFN-ω, IL-27, and IFN-γ (Fig. 4a). PBMCs of patients with AGS harboring mutations in SAMHD1, RNASEH2B, or TREX1 were used as controls. Similar to chronic IFN-I stimulation in vitro [12], tyrosine phosphorylation of residue Y690 on STAT2 (but not pY701-STAT1) was decreased in unstimulated CD14⁺ monocytes of AGS patients with mutations in SAMHD1, RNASEH2B, and TREX1 (Fig. 4b). Decreased STAT2 phosphorylation was not due to impaired IFNAR (interferon-α/β receptor) expression (Fig. 4c). Quantifying mRNA expression of negative regulators of JAK-STAT signaling such as SOCS1, SOCS3, and USP18, we observed significantly increased expression levels of USP18, which correlated with the ISG score (Fig. 4d). Of note, mRNA expression of STAT1 was normal and STAT2 mRNA expression mildly upregulated compared to HCs (Fig. S6). Upon stimulation with IFN-α2 and IFN-ω (IFN-Is; STAT2 dependent), the AGS genotypes (SAMHD1, RNASEH2B, TREX1) with a prominent IFN-I signature in peripheral blood exhibited impaired Y701 phosphorylation of STAT1 (Fig. 4e). In contrast, stimulation with IFN-γ and IL-27 (both STAT2 independent) resulted in increased pY701-STAT1 in CD14⁺ monocytes of patients with mutations in RNU7-1 and other AGS genotypes (Fig. 4f).

Classical AGS presents in the first few months after birth with psychomotor retardation and neurological features including epileptic seizures, motor disorders (dystonia/spasticity), eye movement abnormalities, or spastic paraparesis [3]. To gain further insight into the clinical phenotype associated with mutations in RNU7-1, we incorporated the clinical findings of our 3 cases in the previously described cohort of patients with RNU7-1-related disease (Table 1) [7]. Psychomotor retardation as well as spasticity and dystonia is present in all patients (19/19, 100%). Neuroradiologic imaging revealed intracranial calcifications (16/19, 84%) and leukodystrophy (14/19, 74%) in most patients. In addition, a clinically relevant association with liver damage (10/19, 53%), arterial hypertension (7/19, 37%), pericardial effusions (5/19, 26%), and kidney disease (8/19, 42%), as observed in P3, is found in RNU7-1 patients. Moreover, RNU7-1 patients have a median age of 9 years and the highest mortality rate (8/19, 42%) among AGS genotypes (Fig. S7).